EP1589665A1 - Dekoder zur Fehlerkorrektur mit einem Reed-Solomon Dekoder zur Erasure-Korrektur und einem CRC Dekoder - Google Patents

Dekoder zur Fehlerkorrektur mit einem Reed-Solomon Dekoder zur Erasure-Korrektur und einem CRC Dekoder Download PDF

Info

Publication number
EP1589665A1
EP1589665A1 EP05250926A EP05250926A EP1589665A1 EP 1589665 A1 EP1589665 A1 EP 1589665A1 EP 05250926 A EP05250926 A EP 05250926A EP 05250926 A EP05250926 A EP 05250926A EP 1589665 A1 EP1589665 A1 EP 1589665A1
Authority
EP
European Patent Office
Prior art keywords
data
decoder
blocks
partial
queue
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
EP05250926A
Other languages
English (en)
French (fr)
Other versions
EP1589665B1 (de
Inventor
Matt Ball
Justin Koller
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Quantum Corp
Original Assignee
Quantum Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Quantum Corp filed Critical Quantum Corp
Publication of EP1589665A1 publication Critical patent/EP1589665A1/de
Application granted granted Critical
Publication of EP1589665B1 publication Critical patent/EP1589665B1/de
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/61—Aspects and characteristics of methods and arrangements for error correction or error detection, not provided for otherwise
    • H03M13/615—Use of computational or mathematical techniques
    • H03M13/616—Matrix operations, especially for generator matrices or check matrices, e.g. column or row permutations
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
    • H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
    • H03M13/13—Linear codes
    • H03M13/15—Cyclic codes, i.e. cyclic shifts of codewords produce other codewords, e.g. codes defined by a generator polynomial, Bose-Chaudhuri-Hocquenghem [BCH] codes
    • H03M13/151—Cyclic codes, i.e. cyclic shifts of codewords produce other codewords, e.g. codes defined by a generator polynomial, Bose-Chaudhuri-Hocquenghem [BCH] codes using error location or error correction polynomials
    • H03M13/1515—Reed-Solomon codes
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/29—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes combining two or more codes or code structures, e.g. product codes, generalised product codes, concatenated codes, inner and outer codes
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/29—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes combining two or more codes or code structures, e.g. product codes, generalised product codes, concatenated codes, inner and outer codes
    • H03M13/2906—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes combining two or more codes or code structures, e.g. product codes, generalised product codes, concatenated codes, inner and outer codes using block codes
    • H03M13/2909—Product codes
    • H03M13/2915—Product codes with an error detection code in one dimension
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/37—Decoding methods or techniques, not specific to the particular type of coding provided for in groups H03M13/03 - H03M13/35
    • H03M13/373—Decoding methods or techniques, not specific to the particular type of coding provided for in groups H03M13/03 - H03M13/35 with erasure correction and erasure determination, e.g. for packet loss recovery or setting of erasures for the decoding of Reed-Solomon codes
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/65—Purpose and implementation aspects
    • H03M13/6561—Parallelized implementations

Definitions

  • the present application generally relates to error correction of data using error correction codes (e.g., Bose-Chaudhuri-Hocquenghem (BCH), Reed Solomon codes, and the like), and more particularly to an error correction decoder using cells with partial syndrome generation.
  • error correction codes e.g., Bose-Chaudhuri-Hocquenghem (BCH), Reed Solomon codes, and the like
  • Error correction of data may be used in various applications, such as data storage devices, telecommunication systems, and the like.
  • data is stored by writing the data to a storage medium in the storage device.
  • the stored data can be later retrieved from the storage device by reading the data from the storage medium.
  • an error may exist in the data retrieved from the storage device, meaning that the stored data cannot be retrieved or is different from the data originally stored on the storage medium.
  • a portion of the stored data on the storage medium may degenerate over time such that the portion of the stored data cannot be properly read at a later time.
  • Conventional error correction techniques include generating or encoding one or more redundancy blocks for data, which can be used in a decoding process to correct errors in the data.
  • the decoding process is performed using specialized hardware, which tends to be complicated and difficult to modify.
  • a decoder to correct errors in data includes a plurality of cells. Each cell generates a partial syndrome based on data blocks and one or more redundancy blocks. Each cell generates a partial error value based on a portion of an inverse of an error location matrix that identifies locations of any data blocks having errors. A summation logic connected to the plurality of cells generates an error value based on the partial error values generated by the plurality of cells. The error value corrects errors in a data block having errors.
  • Fig. 1 depicts an exemplary host terminal connected to an exemplary storage device
  • Fig. 2 depicts an exemplary entity having a set of data blocks, redundancy blocks, and cyclic redundancy checking codes
  • FIG. 3 depicts portions of the exemplary entity of Fig. 2;
  • Fig. 4 depicts an exemplary decoder
  • Fig. 5 depicts an exemplary cell of the exemplary decoder depicted in Fig. 4;
  • Fig. 6 depicts an exemplary read process performed by the exemplary cell depicted in Fig. 5;
  • Fig. 7 depicts a portion of the exemplary cell depicted in Fig. 5;
  • Fig. 8 depicts an exemplary write process performed by the exemplary cell depicted in Fig. 5, and
  • Fig. 9 depicts another portion of the exemplary cell depicted in Fig. 5.
  • error correction of data in a storage device is described below. It should be recognized, however, that error correction can be used in various applications, including telecommunications.
  • a host terminal 102 is depicted connected to a storage device 104.
  • Host computer 102 can be any type of computer, such as a personal computer, a workstation, a server, and the like.
  • Storage device 104 can be any type of storage drive, such as a tape drive, a hard drive, and the like. It should be recognized that host terminal 102 can be connected to any number of storage devices 104, and any number of host terminals 102 can be connected to one or more storage devices 104.
  • storage device 104 is configured to detect and correct errors in data stored in storage device 104. More specifically, when data stored in storage device 104 is retrieved, storage device 104 is configured to use redundancy blocks, which are also referred to as error correction code (ECC) redundancy blocks, to correct errors in the retrieved data, such as if the retrieved data is different from the data that was originally stored in storage device 104 or if the stored data cannot be retrieved. Additionally, inner codes, such as cyclic redundancy checking (CRC) codes, can be used to detect errors in the retrieved data. However, it should be recognized that error correction codes, such as Reed-Solomon codes, can be used to detect as well as correct errors.
  • ECC error correction code
  • storage device 104 includes a storage medium 106, a channel and read/write head 108, a processor 110, and an error detection/correction unit 112.
  • data is stored in storage medium 106.
  • Read/write head 108 reads and/or writes data to storage medium 106.
  • Processor 110 controls the operation of storage device 104, including the operation of channel and read/write head 108.
  • error detection/correction unit 112 detects and corrects errors in data stored in storage medium 106.
  • error detection/correction unit 112 includes a data buffer 114, a redundancy block encoder/decoder 116, and an inner code encoder/decoder 118.
  • data is received from host terminal 102 and written to data buffer 114.
  • Redundancy block encoder/decoder 116 generates redundancy blocks for data in data buffer 114.
  • Inner code encoder/decoder 118 generates inner codes (e.g., CRC codes, Reed-Solomon codes, and the like) for data in data buffer 114.
  • Read/write head 108 then writes the data and the generated redundancy blocks and inner codes to storage medium 106.
  • read/write head 108 When data is to be read from storage medium 106, read/write head 108 reads data, redundancy blocks, and inner codes from storage medium 106 to data buffer 114. As will be described in greater detail below, any errors in the data read from storage medium 106 are detected and corrected using the inner codes and redundancy blocks. The data may then be transferred to host terminal 102.
  • data is transferred between host terminal 102 and storage device 104 in data records, which are stored in a buffer.
  • the data records are divided into data blocks of a predetermined length, such as two kBytes, four kBytes, six kBytes, and the like. It should be recognized, however, that data blocks of various lengths may be used.
  • retrieved data blocks having errors are detected, where an error in a retrieved data block indicates that the data in the retrieved data block cannot be read or is different than the data in the data block when the data block was originally stored in storage medium 106.
  • CRC codes can be used to detect when the data in the retrieved data block is different from the data in the data block when the data was originally stored in storage medium 106. More specifically, prior to storing a data block in storage medium 106, a CRC code is generated for the data block and stored with the data block in storage medium 106. When the data block is later retrieved, a new CRC code is generated for the retrieved data block.
  • the new CRC code is then compared to the CRC code retrieved from storage medium 106, which corresponds to the retrieved data block and was originally generated for the retrieved data block before storing the retrieved data block in storage medium 106. If the new CRC code and the retrieved CRC code differ, then an error is detected for that data block. It should be recognized, however, that various types of error detection codes, including Reed-Solomon codes, may be used.
  • an error location matrix identifies the location of any data retrieved from the storage medium having errors.
  • an exemplary error location matrix can be expressed as: where X is the location of an error and p is the total number of data blocks and redundancy blocks in error.
  • the inverse of the error location matrix is generated.
  • the inverse of the exemplary error location matrix in the above example can be expressed as follows: Note that the error location matrix is a Vandermonde matrix, which in general requires O( ⁇ 2 ) operations to compute the inverse, in comparison to general matrix inversions, which requires O( ⁇ 3 ) operations. A row in a Vandermonde matrix consists of a constant value raised to incrementing powers, starting at zero.
  • the inverse of the error location matrix can by generated by firmware or hardware using various matrix inversion techniques.
  • redundancy blocks are used to correct errors in the retrieved data blocks. More specifically, prior to storing data blocks in storage medium 106, redundancy blocks are generated based on the data blocks, and stored with the data blocks in storage medium 106. When the data blocks are later retrieved, data blocks identified as having errors are corrected using the redundancy blocks.
  • redundancy blocks are Bose-Chaudhuri-Hocquenghem (BCH) codes, and more particularly Reed-Soloman codes.
  • BCH Bose-Chaudhuri-Hocquenghem
  • Reed-Solomon codes see Peterson & Weldon, Error Correcting Codes, 2d Edition, MIT Press, 1972, which is incorporated in its entirety herein by reference. It should be recognized, however, that various types of error correction codes may be used.
  • a set of data blocks, a set of redundancy blocks, and a set of redundancy symbols of an inner code are read and written together as a group referred to as an "entity.”
  • an entity 202 is depicted having 16 data blocks 204, four redundancy blocks 206, and 20 redundancy symbols of an inner code 208.
  • entity 202 can include various numbers of data blocks 204, redundancy blocks 206, and redundancy symbols of an inner code 208.
  • entity 202 can include 32 data blocks 204 and eight redundancy blocks 206, 112 data blocks 204 and 16 redundancy blocks 206, and the like.
  • redundancy symbols of an inner code 208 can both detect and correct error within a data block 204 or redundancy block 206.
  • Fig. 2 depicts the form in which entity 202 can be stored in data buffer 114 (Fig. 1). It should be recognized, however that entity 202 need not exist physically in the form depicted in Fig. 2. It should also be recognized that data in entity 202, and more particularly the data in a data block 204, need not correspond to a single file. Instead, in the present exemplary embodiment, data received from host terminal 102 (Fig. 1) is interleaved. As such, the data in a particular data block 204 can correspond to portions of separate files received from host terminal 102 (Fig. 1).
  • Fig. 2 also depicts logical relationships between data blocks 204, redundancy blocks 206, and redundancy symbols of an inner code 208 of entity 202. With reference to Fig. 3, portions of entity 202 are shown in greater detail to more clearly illustrate the logical relationships between data blocks 204, redundancy blocks 206, and redundancy symbols of an inner code 208.
  • redundancy symbols of an inner code 208 are depicted as being CRC codes. It should be recognized, however, that various types of error detection or error correction codes can be used, such as Reed-Solomon codes.
  • redundancy symbols of an inner code 208 corresponds to a data block 204 or redundancy block 206 and is used to detect an error in data block 204 or redundancy block 206.
  • CRC code CRC 19 corresponds to data block D 19 of entity 202.
  • a new CRC code CRC 19 ' is generated for retrieved data block D 19 .
  • the new CRC code CRC 19 ' is then compared to the CRC code retrieved from storage medium 106 (Fig. 1) corresponding to retrieved data block D 19 (i.e., CRC code CRC 19 ). If the new CRC code CRC 19 ' and the retrieved CRC code CRC 19 differ, then an error is detected for data block D 19 .
  • the number of redundancy blocks determines the maximum number of data blocks that can be corrected.
  • a total of four redundancy blocks 206 may be used to correct a maximum of four data blocks 204 in error.
  • each redundancy block 204 is generated based on the data in all of the data blocks of entity 202.
  • redundancy blocks E 0 , E 1 , E 2 , and E 3 are each generated based on the data in data blocks D 4 , D 2 , ..., and D 19 .
  • redundancy blocks can be generated by redundancy block encoder/decoder 116.
  • redundancy blocks are initially generated for data received from host terminal 102. The generated redundancy blocks and the received data are then stored in storage medium 106.
  • redundancy blocks E 0 , E 1 , E 2 , and E 3 are generated based on the same set of data (i.e., data blocks 204 of entity 202), each redundancy block 206 is unique as to each other. More specifically, in the present embodiment, redundancy blocks E 0 , E 1 , E 2 , and E 3 are Bose-Chaudhuri-Hocquenghem (BCH) codes, and more particularly Reed-Solomon codes.
  • BCH Bose-Chaudhuri-Hocquenghem
  • redundancy block encoder/decoder 116 operates as an encoder to generate redundancy blocks to be stored with data blocks in storage medium 106. Redundancy block encoder/decoder 116 operates as a decoder to correct errors in data blocks retrieved from storage medium 106. It should be recognized, however, that redundancy block encoder/decoder 116 can be implemented as separate components (i.e., an encoder component and a decoder component) in storage device 104.
  • encoder 400 can be implemented as an integrated part of redundancy block encoder/decoder 116 (Fig. 1) or as a separate component within storage device 104 (Fig. 1).
  • encoder 400 includes a plurality of cells 402 and a summation logic 404 connected to the plurality of cells 402.
  • Each cell 402 generates a partial syndrome based on the data blocks and redundancy blocks retrieved from the storage medium.
  • Each cell 402 also generates a partial error value based on a portion of the inverse of an error location matrix that identifies the locations of any data blocks retrieved from the storage medium having errors.
  • the inverse of the error location matrix can be generated by firmware or hardware.
  • Summation logic 404 generates an error value, which corrects errors in a data block retrieved from the storage medium, based on the partial error values generated by plurality of cells 402.
  • data blocks and redundancy blocks retrieved from storage medium 106 are held in data buffer 114. More particularly, with reference to Fig. 2, in one exemplary embodiment, data blocks 204 and redundancy blocks 206 can be stored in data buffer 114 (Fig. 1) in the form of entity 202.
  • plurality of cells 402 are connected to data buffer 114 (Fig. 1) through lines 406.
  • Each cell 402 receives as an input a portion of entity 202 (Fig. 2) stored in data buffer 114 (Fig. 1). More particularly, each cell 402 reads in portions of entity 202 (Fig. 2) in cache bursts.
  • each cache burst is a portion of a single data block 204 or redundancy block 206. Additionally, in the present exemplary embodiment, portions of data blocks 204 and redundancy blocks 206 are read in a raster pattern.
  • each cache burst is 32 bytes long.
  • the first 32 bytes of the first data block 204 is read in a first cache burst.
  • the first 32 bytes of each of the next 15 data blocks 204 and four redundancy blocks 206 of entity 202 are then read in 19 subsequent cache bursts.
  • the second subsequent 32 bytes of the first data block 204 is read, and then the second subsequent 32 bytes of each of the next 15 data blocks 204 and four redundancy blocks 206 of entity 202 are read.
  • data blocks 204 and redundancy blocks 206 of entity 202 are read in cache bursts in a raster pattern in 32 byte portions.
  • the size of the cache burst can vary, and data blocks 204 and redundancy blocks 206 can be read in various patterns.
  • each cache burst can be 64 bytes rather than 32 bytes.
  • a portion of each data block 204 and redundancy block 206 in entity 202 can be logically grouped as a code word 210.
  • the first portions of each data block 204 and redundancy block 206 can be logically grouped as a first code word 210, which corresponds to the first column in Fig. 2.
  • the second byte of each data block 204 and redundancy block 206 can be logically grouped as a second code word 210, which corresponds to the second column in Fig. 2.
  • code word 210 is one byte wide.
  • data blocks 204 and redundancy blocks 206 are 2 kBytes long, then data blocks 204 and redundancy blocks 206 in entity 202 can be logically grouped into 2,000 separate, individual code words 210.
  • data blocks 204 and redundancy blocks 206 are read using a cache burst of 32 bytes, which corresponds to 16 long words of 32 bits each, then portions of 32 code words 210 are read at a time.
  • the number of cells 402 in decoder 400 corresponds to the number of redundancy blocks 206 (Fig. 2) in entity 202, which in turn corresponds to the maximum number of data blocks 204 that can be corrected.
  • decoder 400 can be modified based on the maximum number of data blocks 204 to be corrected. For example, when a maximum of four data blocks 204 are to be corrected, decoder 400 is modified to include four cells 402. Similarly, when a maximum of 16 or 32 data blocks 204 are to be corrected, decoder 400 is modified to include 16 or 32 cells 402.
  • each cell 402 includes an input logic 502, a queue 504, a multiplier 506, a multiplexer 508, and queues 510 and 512.
  • each cell 402 generates a partial syndrome based on data blocks and one or more redundancy blocks retrieved from the storage medium.
  • FIG. 6 an exemplary read process 600 of generating a partial syndrome in cell 400 (Fig. 4) is depicted.
  • cell 402 (Fig. 4) is set to a read mode.
  • input logic 502 is set to receive inputs from line 406, which is connected to data buffer 114 (Fig. 1), and feedback line 514, which is connected to the output of multiplier 506.
  • multiplexer 702 is set to receive input from an XOR gate 704 connected to line 406 and feedback line 514.
  • multiplier 506 is set to receive inputs from queue 504, which holds intermediate results generated by input logic 502, and queue 510, which holds a root of a generator polynomial.
  • multiplexer 508 with inputs connected to queues 510 and 512 is set to receive input from queue 510.
  • cell 402 reads data from data buffer 114 (Fig. 1).
  • cell 402 reads in a portion of a data block or a redundancy block from entity 202 (Fig. 2) stored in data buffer 114 (Fig. 1) as a cache burst.
  • data can be read from data buffer 114 (Fig. 1) in a cache burst of 32 bytes, which corresponds to 16 long words.
  • line 406 and feedback line 514 are 32 bits wide, which allows cell 402 to process 32 bits (one long word) at a time. It should be recognized, however, that lines 406 and feedback line 514 can be any size, and cell 402 can process any number of bits at a time.
  • input logic 502 performs an exclusive-OR (XOR) operation on a cache burst read from entity 202 (Fig. 2) through line 406 and the output of multiplier 506 through feedback line 514.
  • input logic 502 includes XOR gate 704, which can perform the XOR operation. It should be recognized, however, that input logic 502 can include various components, including various types and numbers of logic gates, to sum the data read from data buffer 114 (Fig. 1) with the output from multipliers 504 (Fig. 5).
  • the sum of the data read from data buffer 114 (Fig. 1) and the output from multiplier 506 (Fig. 5) is stored as an intermediate result.
  • the intermediate result generated by input logic 502 is stored in queue 504.
  • the size and number of entries in queue 504 is determined based on the size of the cache burst used to read data from data buffer 114 (Fig. 1) and the size of line 406 and feedback line 514. For example, if data is read from data buffer 114 (Fig. 1) in a cache burst of 32 bytes, which corresponds to 16 long words, and line 516 and feedback line 514 are 32 bits wide, then each long word in the cache burst is summed with the output from multiplier 506 and stored as an entry in queue 504. Thus, in this example, queue 504 includes eight entries with each entry 32 bits long. If the size of the cache burst is changed, then the size of queue 504 can also be changed. For example, if the cache burst is 64 bytes long, then queue 504 can include 16 entries with each entry 32 bits long.
  • multiplier 506 is a Galois Field Multiplier that performs a Galois Field multiplication between the intermediate result stored in queue 504 and a root of a generator polynomial stored in queue 510.
  • multiplier 506 performs multiple, parallel Galois Field multiplications. For example, if each entry of queue 504 is 32 bits long, the 32 bits of data from an entry of queue 504 can be formatted as four parallel eight bit Galois Field elements. Multiplier 506 can then perform four parallel eight-by-eight Galois Field multiplications between the 32 bit long entry from queue 504 and an eight bit root of a generator polynomial from queue 510.
  • loop 604 to 612 is iterated to process the next data block or redundancy block.
  • the next data block to be processed was determined to have errors, then in 604 all zeros are read rather than reading a cache burst from data buffer 114 (Fig. 1). If all of the data blocks and redundancy blocks have been processed, read process 600 terminates at 614.
  • the number of iterations of loop 604 to 612 corresponds to the number of data blocks and redundancy blocks in entity 202 (Fig. 2).
  • Fig. 2 For example, with reference to Fig. 3, assume that portions of entity 202 are read in cache bursts of 32 bytes in length.
  • a first cache burst containing the first 32 bytes of data block D 19 is read and processed.
  • a second cache burst containing the first 32 bytes of data block D 18 is read and processed. Assume that data block D 4 was identified as having errors.
  • loop 604 to 612 (Fig. 6)
  • all zeros are read and processed rather than reading in a cache burst from data block D 4 .
  • loop 604 to 612 (Fig. 6) is iterated 20 times to process all 16 data blocks and four redundancy blocks.
  • the result of the sum of the second cache burst and the output of the multiplier 506 is the sum of the data read in the second cache burst and the result of multiplying the data read in the first cache burst by a root of a generator polynomial.
  • This result is then stored in queue 504 as the new intermediate result.
  • the final result stored in queue 504 of each cell 402 is the partial syndrome for each cell 402.
  • each cell 402 uses a different root of a generator polynomial.
  • different partial syndromes are generated in each of the plurality of cells 402.
  • each cell 402 uses a root of a generator polynomial of ⁇ ", where n is the sequence number of the cell.
  • n is the sequence number of the cell.
  • cells 402 can be assigned sequence numbers from zero to ( ⁇ -1) (i.e., cell 402(0), cell 402(1), cell 402(2), ..., cell 402 ( ⁇ -1)).
  • cell 402(0) uses ⁇ 0
  • cell 402(1) uses ⁇ 1
  • cell 402(2) uses ⁇ 2
  • ... cell 402( ⁇ -1) uses ⁇ ( ⁇ -1 )
  • the root of a generator polynomial can be a hardwired constant, or a register set by firmware.
  • each cell 402 can use a root of a generator polynomial of ⁇ n+1 , where n is the sequence number of the cell and l is a fixed constant for each cell 402.
  • each cell 402 uses a different root of a generator polynomial, after process 600 (Fig. 6) is terminated, when all of the data blocks and redundancy blocks in entity 202 (Fig. 2) have been processed, different partial syndromes have been generated by each cell 402 and stored in the queues 504 of cells 402. For example, if there are p cells 402 and cells 402 are assigned sequence numbers from zero to ( ⁇ -1) (i.e., cell 402(0), cell 402(1), cell 402(2), ..., cell 402 ( ⁇ -1)), cell 402(0) generates partial syndrome So, cell 402(1) generates partial syndrome S 1 , ..., and cell 402( ⁇ -1) generates partial syndrome S( ⁇ -1).
  • each cell 402 After read process 600 (Fig. 6) has been terminated, each cell 402 generates a partial error value based on the generated partial syndrome and a portion of the inverse of the error location matrix.
  • an exemplary write process 800 of generating a partial error value in cell 402 (Fig. 4) is depicted.
  • cell 402 (Fig. 4) is set to a write mode.
  • input logic 502 is set to receive inputs from feedback line 516, which is connected to queue 504, which holds the partial syndrome generated in cell 400 during the previous read process.
  • multiplexer 702 is set to receive input from feedback line 516.
  • multiplier 506 is set to receive inputs from queue 504 and queue 512, which holds a portion of the inverse of the error location matrix.
  • multiplexer 508 with inputs connected to queues 510 and 512 is set to receive input from queue 512.
  • an error location matrix identifies the locations of any data blocks retrieved from the storage medium having errors.
  • an exemplary error location matrix can be expressed as: where X is the location of an error and ⁇ is the total number of data blocks and redundancy blocks in error.
  • the inverse of the error location matrix can be expressed as follows: As also described above, in the present exemplary embodiment, the inverse of the error location matrix can be generated by firmware or hardware.
  • each cell 402 uses a different portion of the inverse of the error location matrix. For example, with reference to Fig. 4, if there are ⁇ cells 402 assigned sequence numbers from zero to ( ⁇ -1) (i.e., cell 402(0), cell 402(1), cell 402(2), ..., cell 402 ( ⁇ -1)), cell 402(0) uses the first row of the inverse of the error location matrix (i.e., U 0,0 , U 1,0 , ...,U ⁇ - 1 ,0 ), cell 402(1) uses the second row of the inverse of the error location matrix (i.e., U 0,1 , U 1,1 , ..., U ⁇ - 1 ,1 ), ..., and cell 402( ⁇ -1) uses the ( ⁇ -1) row of the inverse of the error location matrix (i.e., U 0, ⁇ - 1 , U 1, ⁇ - 1 , ..., U ⁇ - 1 , ⁇ - 1 ).
  • multiplier 506 generates a partial error value by multiplying the partial syndrome, which is held in queue 504, with an element of the portion of the inverse of the error location matrix, which is held in queue 512.
  • ⁇ cells 402 assigned sequence numbers from zero to ( ⁇ -1) (i.e., cell 402(0), cell 402(1), cell 402(2), ..., cell 402 ( ⁇ -1)).
  • a first partial error value y 0 (0) can be generated by multiplying partial syndrome So with the first element from the first row of the inverse of the error location matrix U 0,0 .
  • a second partial error value y 0 (1) can be generated by multiplying partial syndrome S 1 with the first element from the second row of the inverse of the error location matrix U 0,1 .
  • a ( ⁇ -1) partial error value y 0 ( ⁇ -1) can be generated by multiplying partial syndrome S ( ⁇ -1) with the first element from the ( ⁇ -1 ) row of the inverse of the error location matrix U 0, ⁇ -1 .
  • the partial error value generated by cell 402 (Fig. 4) is sent to summation logic 404 (Fig. 4).
  • the partial error values y 0 (0), y 0 (1), ..., and y 0 ( ⁇ -1) generated by cells 402(0), 402(1), ..., and 402( ⁇ -1) are sent to summation logic 404.
  • summation logic 404 sums the partial error values generated by cells 402.
  • a first error value Y 0 is generated by summing the partial error values y 0 (0), y 0 (1), ..., and y 0 ( ⁇ -1) generated by cells 402(0), 402(1), ..., and 402( ⁇ -1).
  • summation logic 404 can include an array of XOR gates 902 that computes the Galois Field sum of the outputs of cells 402 (Fig. 4). It should be recognized, however, that summation logic 404 can include various components, including various types and numbers of logic components, to compute the Galois Field sum of the outputs of cells 402 (Fig.4).
  • the generated error value can then be written to data buffer 114 (Fig. 1) to correct the data in the location corresponding to the elements of a portion of the inverse of the error location matrix used to generate the partial error values in cells 402.
  • partial error values y 0 (0), y 0 (1), ..., and y 0 ( ⁇ -1) are generated in cells 402(0), 402(1), ..., and 402( ⁇ -1) using elements U 0,0 , U 0,1 , ..., and U 0, ⁇ -1 , of a portion of the inverse of the error location matrix.
  • Error value Y 0 can then be written to data buffer 114 (Fig. 1) to correct the portion of data block D 4 .
  • cell 402 when the size of the cache burst is greater than the amount of data processed at a time by cell 402, during write process 800 (Fig. 8), cell 402 cycles through the cache burst to process the entire cache burst.
  • cells 402 can receive data from data buffer 114 (Fig. 1) in a cache burst of 32 bytes and process one long word of the cache burst at a time.
  • queue 504 (Fig. 5) can include eight entries that are each 32 bits (one long word) in length.
  • multiplier 506 in cell 402(0) multiplies each entry in queue 504 (Fig.
  • Summation logic 404 also generates eight sets of error values Y 0 , which are then written as a cache burst back to data buffer 114 (Fig. 1).
  • Fig. 1 For example, with reference to Fig. 3, assume elements U 0,0 , U 0,1 , ..., and U 0, ⁇ -1 correspond to the first 32 bytes of data block D 4 , which corresponds to the first 32 code words of data block D 4 , the eights sets of error values Y 0 are written to the correct the first 32 bytes of data block D 4 .
  • write process 800 (Fig. 8) can then be iterated to generate another set of partial error values, which can be summed to generate another error value, which can then be used to correct a portion of another data block with errors. For example, in a second iteration, a second set of partial error values are generated and summed to generate a second error value.
  • a first partial error value y 1 (0) can be generated by multiplying partial syndrome So with the second element from the first row of the inverse of the error location matrix U 1,0 .
  • a second partial error value y 1 (1) can be generated by multiplying partial syndrome S 1 with the second element from the second row of the inverse of the error location matrix U 1,1 .
  • a ( ⁇ -1) partial error value y 0 ( ⁇ -1) can be generated by multiplying partial syndrome S ( ⁇ -1) with the second element from the ( ⁇ -1) row of the inverse of the error location matrix U 1, ⁇ -1 .
  • the second set of partial error values y 1 (0), y 1 (1), ..., and y 1 ( ⁇ -1) generated by cells 402(0), 402(1), ..., and 402( ⁇ -1) are sent to summation logic 404.
  • a second error value Y 1 is then generated by summing the partial error values y 1 (0), y 1 (1), ..., and y 1 ( ⁇ -1).
  • error values can be generated to correct portions the data blocks identified as having error.
  • the number of times write process 800 (Fig. 8) is iterated and the number of error values generated can be determined based on the number of data blocks identified as having errors (i.e., p).
  • the number of times write process 800 (Fig. 8) is iterated and the number of error values generated can be set to the maximum number of data blocks that can be corrected, which corresponds to the number of redundancy blocks in entity 202 (Fig. 2).
  • read process 600 (Fig. 6) can be iterated again to process another portion of the data blocks and redundancy blocks stored in data buffer 114 (Fig. 1), and then write process 800 (Fig. 8) can be iterated again to correct errors in another portion of the data blocks having errors.
  • queue 504 (Fig. 5) is cleared.
  • Read process 600 (Fig. 6) is iterated to read and process the first 32 bytes of each data block 204 and each redundancy block 206 to generate partial syndromes.
  • Write process 800 (Fig. 8) is then iterated to generate error values and correct errors in the first 32 bytes of any data blocks 204 having errors.
  • Queue 504 (Fig. 5) is cleared, then read process 600 (Fig. 6) is iterated again to read and process the second 32 bytes of each data block 204 and each redundancy block 206 to generate another set of partial syndromes.
  • Write process 800 (Fig.
  • decoder 4 has been described in connection with a storage device, it should be recognized that decoder 4 can be used in connection with various devices and in various applications where data is received through a channel that introduces error, such as part of a telecommunication system. Therefore, the present invention should not be construed as being limited to the specific forms shown in the drawings and described above.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Probability & Statistics with Applications (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Algebra (AREA)
  • Mathematical Optimization (AREA)
  • Computational Mathematics (AREA)
  • Computing Systems (AREA)
  • Detection And Correction Of Errors (AREA)
  • Error Detection And Correction (AREA)
  • Techniques For Improving Reliability Of Storages (AREA)
EP05250926A 2004-02-19 2005-02-18 Dekoder zur Fehlerkorrektur mit einem Reed-Solomon Dekoder zur Erasure-Korrektur und einem CRC Dekoder Expired - Fee Related EP1589665B1 (de)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US782990 1991-10-28
US10/782,990 US7228490B2 (en) 2004-02-19 2004-02-19 Error correction decoder using cells with partial syndrome generation

Publications (2)

Publication Number Publication Date
EP1589665A1 true EP1589665A1 (de) 2005-10-26
EP1589665B1 EP1589665B1 (de) 2006-11-15

Family

ID=34861115

Family Applications (1)

Application Number Title Priority Date Filing Date
EP05250926A Expired - Fee Related EP1589665B1 (de) 2004-02-19 2005-02-18 Dekoder zur Fehlerkorrektur mit einem Reed-Solomon Dekoder zur Erasure-Korrektur und einem CRC Dekoder

Country Status (4)

Country Link
US (1) US7228490B2 (de)
EP (1) EP1589665B1 (de)
JP (1) JP4733403B2 (de)
DE (1) DE602005000251T2 (de)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7431585B2 (en) * 2002-01-24 2008-10-07 Applied Materials, Inc. Apparatus and method for heating substrates
US7426678B1 (en) * 2004-07-20 2008-09-16 Xilinx, Inc. Error checking parity and syndrome of a block of data with relocated parity bits
US7694208B2 (en) * 2005-12-20 2010-04-06 Quantum Corporation Error correction algorithm using interleaved parity check and Reed-Solomon code
KR100833600B1 (ko) * 2006-08-25 2008-05-30 삼성전자주식회사 에러 정정 회로, 그 방법 및 상기 회로를 구비하는 반도체메모리 장치
US8271857B2 (en) * 2008-05-13 2012-09-18 International Business Machines Corporation Correcting errors in longitudinal position (LPOS) words
US9179362B2 (en) * 2011-08-25 2015-11-03 Texas Instruments Incorporated Systems and methods for networking coding using Reed-Solomon codes
US10181864B2 (en) * 2016-02-26 2019-01-15 Altera Corporation Methods and apparatus for performing reed-solomon encoding
JP7400514B2 (ja) * 2020-02-07 2023-12-19 Tdk株式会社 メモリコントローラ及びフラッシュメモリシステム

Family Cites Families (41)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4646303A (en) * 1983-10-05 1987-02-24 Nippon Gakki Seizo Kabushiki Kaisha Data error detection and correction circuit
JPH0812612B2 (ja) * 1983-10-31 1996-02-07 株式会社日立製作所 誤り訂正方法及び装置
JPH0770177B2 (ja) * 1984-01-25 1995-07-31 株式会社日立製作所 ディジタル信号再生装置
NL8400630A (nl) * 1984-02-29 1985-09-16 Philips Nv Decodeerinrichting voor een stroom van codesymbolen die woordsgewijze beschermd zijn door een dubbele reed-solomon-code met een minimum hamming-afstand van 5 over de codesymbolen en een verbladeringsmechanisme tussen de beide codes, alsmede speler voorzien van zo een decodeerinrichting.
US4777635A (en) * 1986-08-08 1988-10-11 Data Systems Technology Corp. Reed-Solomon code encoder and syndrome generator circuit
FR2605769B1 (fr) * 1986-10-22 1988-12-09 Thomson Csf Operateur polynomial dans les corps de galois et processeur de traitement de signal numerique comportant un tel operateur
US5325373A (en) * 1986-12-22 1994-06-28 Canon Kabushiki Kaisha Apparatus for encoding and decoding reed-solomon code
JPS63193723A (ja) * 1987-02-06 1988-08-11 Sony Corp リ−ドソロモン符号の復号方法
US5136592A (en) * 1989-06-28 1992-08-04 Digital Equipment Corporation Error detection and correction system for long burst errors
US5430739A (en) * 1990-03-27 1995-07-04 National Science Council Real-time Reed-Solomon decoder
JP2919633B2 (ja) * 1991-05-10 1999-07-12 三井化学株式会社 プロピレンの部分酸化物の製造方法
JP2662472B2 (ja) * 1991-06-13 1997-10-15 シャープ株式会社 誤り訂正処理用シンドローム演算回路
US5285455A (en) * 1992-02-03 1994-02-08 Lsi Logic Corporation Serial data encoder
US5471485A (en) * 1992-11-24 1995-11-28 Lsi Logic Corporation Reed-solomon decoder using discrete time delay in power sum computation
US5444719A (en) * 1993-01-26 1995-08-22 International Business Machines Corporation Adjustable error-correction composite Reed-Solomon encoder/syndrome generator
US5383204A (en) * 1993-06-29 1995-01-17 Mitsubishi Semiconductor America, Inc. Parallel encoding apparatus and method implementing cyclic redundancy check and Reed-Solomon codes
US5771244A (en) * 1994-03-09 1998-06-23 University Of Southern California Universal Reed-Solomon coder/encoder
FR2721774B1 (fr) * 1994-06-27 1996-09-06 Sgs Thomson Microelectronics Décodeur reed-solomon.
US5555250A (en) * 1994-10-14 1996-09-10 Compaq Computer Corporation Data error detection and correction system
JP2963018B2 (ja) * 1994-11-08 1999-10-12 旭化成マイクロシステム株式会社 リード・ソロモン誤り訂正符号復号化回路
KR970706573A (ko) * 1995-07-19 1997-11-03 롤페스 제이. 지. 에이. 멀티프레임 에러 방지 기능을 가진 가역 멀티트랙 자기 테이프 기억장치 시스템(reversible multintrack magnetic tape storage system with multiframe error protection facility)
US5719884A (en) * 1995-07-27 1998-02-17 Hewlett-Packard Company Error correction method and apparatus based on two-dimensional code array with reduced redundancy
US5889792A (en) * 1996-10-01 1999-03-30 Intel Corporation Method and apparatus for generating syndromes associated with a block of data that employs re-encoding the block of data
GB9622539D0 (en) * 1996-10-30 1997-01-08 Discovision Ass Galois field multiplier for reed-solomon decoder
US5948117A (en) * 1997-01-23 1999-09-07 Quantum Corporation Modified Reed-Solomon error correction system using (W+i+1)-bit representations of symbols of GF(2w+i)
US5901158A (en) * 1997-04-22 1999-05-04 Quantum Corporation Error correction encoder/decoder
JP3850511B2 (ja) * 1997-05-07 2006-11-29 日本テキサス・インスツルメンツ株式会社 リードソロモン復号装置
JPH10322226A (ja) * 1997-05-14 1998-12-04 Texas Instr Japan Ltd リードソロモン復号方法
US5889793A (en) * 1997-06-27 1999-03-30 Integrated Device Technology, Inc. Methods and apparatus for error correction
JPH1131977A (ja) * 1997-07-10 1999-02-02 Sony Corp 誤り訂正符号演算器
KR100256250B1 (ko) * 1997-08-30 2000-05-15 김영환 직렬화된 리드 솔리몬 복호장치
US6041431A (en) * 1997-09-19 2000-03-21 Adapter, Inc. Method and apparatus for performing error correction code operations
JPH11196006A (ja) * 1997-12-26 1999-07-21 Nec Corp 並列処理シンドロ−ム計算回路及びリ−ド・ソロモン複合化回路
US6327690B1 (en) * 1999-02-04 2001-12-04 Intel Corporation Integrated reed-solomon error correction code encoder and syndrome generator
US6347389B1 (en) * 1999-03-23 2002-02-12 Storage Technology Corporation Pipelined high speed reed-solomon error/erasure decoder
US6341362B1 (en) * 1999-04-22 2002-01-22 Vlsi Technology, Inc. Extended symbol Galois field error correcting device
JP3502583B2 (ja) * 1999-10-25 2004-03-02 松下電器産業株式会社 誤り訂正方法および誤り訂正装置
US6651208B1 (en) * 2000-04-04 2003-11-18 Mosel Vitelic Corporation Method and system for multiple column syndrome generation
GB0031436D0 (en) * 2000-12-22 2001-02-07 Koninkl Philips Electronics Nv Method and apparatus for data reproduction
JP3954803B2 (ja) * 2001-03-22 2007-08-08 三洋電機株式会社 エラー訂正装置
JP3879082B2 (ja) * 2002-05-31 2007-02-07 財団法人理工学振興会 バイト誤り訂正・検出装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
YOUSHI XU ET AL: "Variable Shortened-and-Punctured Reed-Solomon Codes for Packet Loss Protection", IEEE TRANSACTIONS ON BROADCASTING, IEEE INC. NEW YORK, US, vol. 48, no. 3, September 2002 (2002-09-01), pages 237 - 245, XP002287273, ISSN: 0018-9316 *

Also Published As

Publication number Publication date
EP1589665B1 (de) 2006-11-15
US7228490B2 (en) 2007-06-05
DE602005000251D1 (de) 2006-12-28
JP4733403B2 (ja) 2011-07-27
DE602005000251T2 (de) 2007-06-14
US20050188293A1 (en) 2005-08-25
JP2005293557A (ja) 2005-10-20

Similar Documents

Publication Publication Date Title
US6615387B1 (en) Method and apparatus for error detection
US4706250A (en) Method and apparatus for correcting multibyte errors having improved two-level code structure
US6374383B1 (en) Determining error locations using error correction codes
US6560747B1 (en) Error counting mechanism
US5659557A (en) Reed-Solomon code system employing k-bit serial techniques for encoding and burst error trapping
US9535788B2 (en) High-performance ECC decoder
US7562283B2 (en) Systems and methods for error correction using binary coded hexidecimal or hamming decoding
US4525838A (en) Multibyte error correcting system involving a two-level code structure
US4504948A (en) Syndrome processing unit for multibyte error correcting systems
EP1589665B1 (de) Dekoder zur Fehlerkorrektur mit einem Reed-Solomon Dekoder zur Erasure-Korrektur und einem CRC Dekoder
EP0373764B1 (de) Korrektur von Burst- und Zufallsfehlern
US6279137B1 (en) System and method for a storage-efficient parallel Chien Search
US8201061B2 (en) Decoding error correction codes using a modular single recursion implementation
EP0753942A2 (de) Wortweise Verarbeitung für Reed-Solomon-Codes
US7290197B2 (en) Correcting data using redundancy blocks
US6643819B1 (en) Hybrid root-finding technique
EP0329775B1 (de) Schaltung mit grosser bandbreite und verfahren zur reed-solomon-kodierung, dekodierung und fehlerkorrektur
US7228467B2 (en) Correcting data having more data blocks with errors than redundancy blocks
EP0341851A2 (de) Verfahren und Gerät für verschachtelte Kodierung
US6341362B1 (en) Extended symbol Galois field error correcting device
KR101226439B1 (ko) 리드-솔로몬 디코더, 이를 포함하는 메모리 시스템 및 디코딩 방법

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU MC NL PL PT RO SE SI SK TR

AX Request for extension of the european patent

Extension state: AL BA HR LV MK YU

GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

17P Request for examination filed

Effective date: 20060324

AKX Designation fees paid

Designated state(s): DE GB

GRAS Grant fee paid

Free format text: ORIGINAL CODE: EPIDOSNIGR3

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): DE GB

REG Reference to a national code

Ref country code: GB

Ref legal event code: FG4D

REF Corresponds to:

Ref document number: 602005000251

Country of ref document: DE

Date of ref document: 20061228

Kind code of ref document: P

PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed

Effective date: 20070817

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20070901

GBPC Gb: european patent ceased through non-payment of renewal fee

Effective date: 20090218

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: GB

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20090218